I was just wondering... With all the data this system must collect and the time put in to make the ranking system, if anyone has used the data to see if there is a correlation between the color someone selects and the % of wins. I would think that a brighter color might have an effect on some games. I would assume it depends on how many opponents/players are in a game as well.
There are many variables that would have to be taken in consideration to do it correctly.
Anyway....
I don't think anyone has done that. I believe you could pull the data yourself using the WarGear API and try it figure it out. Sounds interesting. I'd really like to see this for maps where color also is tied to the starting position (e.g. Invention).
This could be done, but unless you are thinking of something pretty sophisticated, only for those maps where standard colors are used. I customize the colors of most of my maps. Also, for those maps where less than 16 players are allowed, often the designer will delete colors, so though you can still calculate win percentages, you'd probably have to use some kind of H-rating related formula to make them relevant and relatable.
For maps like Invention (where apparently color is tied to position), call me an old fashioned enemy of the scientific method, but I'm thinking it's more likely that the position is affecting the win % than the color. Other examples, Waterloo and Go-Geared. where in both cases there are only two colors and they are assigned.
>I'm thinking it's more likely that the position is affecting the win % than the color.
Oh yeah, of course! I just think it would be very interesting to see how the position affects win %. More interesting to me than whether bright colors have a slight advantage/disadvantage to them.
As for the colors, my guess is that a neutral looking grey or a black or something is going to be the only color with any significant advantage (due to player confusion), and even then it will be small.
There have been a number of great discussions about which aspects of games could be analyzed for win % etc. For instance, it would be interesting to see how much of a higher likelihood you have of winning starting with Western Australia over Japan.
I would assume that (not counting games like Invention/Go where they are assigned) that the colors with the most wins would be the colors preferred by the top players.
I was thinking, to get an idea of where to start on this a few basic scatter plots would need to be made. One set of scatter plots would compare all wins on all maps with vs. the color the winner was, then compare all maps with the -actual statistical outlier players- removed (top players and never won players and so keep inexperience players from biasing the results say... less then 10 games played excluded as well). You may already see a correlation at this point. If not then one more set of scatter plots should be made to deduce the idea once and for all.
Next would compare either top played boards or the most basic boards (risk style boards only) and all wins and the color or winner. This may have to be a bar graph by board and by color.... to get a visual...
If a statistical correlation does exist it would show up in one if not all of the scatter plots. Then the more detailed bar graphs/Histograms would need to be made to compare the board and colors to eliminate better colors on what boards.... Wow... This gets pretty complicated fast. But the first set of scatter plots would not be too hard... And if not correlation shows on the first set of scatter plots then the second half would not need to be done at all....
Hope this makes sense.... Sort of thinking out loud here...
Crandawg wrote:One set of scatter plots would compare all wins on all maps with vs. the color the winner was,
To clarify, I mean each color would need its own scatter plot of all win's of all maps..
Then eliminate the best and worst players (outliers) to see is a correlation shows or disappears vs. the All players all Boards...
Korrun wrote:I would assume that (not counting games like Invention/Go where they are assigned) that the colors with the most wins would be the colors preferred by the top players.
Winner.
asm wrote:Korrun wrote:I would assume that (not counting games like Invention/Go where they are assigned) that the colors with the most wins would be the colors preferred by the top players.
Winner.
..or is it the colors themselves that make the winners winners? Cause and effect? Do people pick on cyan more than they pick on pink?
Is that why I keep losing? I love cyan.
I play maroon whenever it's an option.
..don't go there!
I went there.
If you do this analysis, try to factor in the size of the game too. A sixteen player game is likely to see higher survival rates for a harder to see color. A two player game will not.
asm wrote:Korrun wrote:I would assume that (not counting games like Invention/Go where they are assigned) that the colors with the most wins would be the colors preferred by the top players.
Winner.
+1